Analyzing and Testing MAX

Transcription

1 AIQ Opening Bell Newsletter May 2006 Vol. 15 Issue 5 In This Issue Trading pro Steve Palmquist analyzes the Moving Average Crossover System... 1 Learn how you can broaden EDS trading systems by manipulating the pre-built rules... 6 Market Review... 9 Data Maintenance... 9 S&P 500 Changes... 9 Check out AIQ's one-day seminar scheduled in San Diego May The Opening Bell Newsletter is a publication of AIQ Systems P.O. Box 7530 Incline Village, Nevada openingbell@aiqsystems.com Analyzing and Testing MAX Moving Average Crossover System (MAX) Is a Promising Tool for Trader's Tool Box By Steve Palmquist Steve Palmquist is the founder of which provides free market analysis, trading tips, and educational material. A fulltime trader with 20 years of experience, he is a featured speaker at AIQ seminars. O ne of the key observations that helped improve my trading was that the market alternates between several different environments, and each environment has trading techniques that work and ones that don t. This observation led me to develop specific tools for each type of environment, and techniques for determining which tool to use. Having the right set of tools and knowing which one to use for a specific task is one of the keys to success in trading, as it is in many other professions. Carpenters have more than one tool in their toolboxes, pilots use multiple instruments for navigation; and traders need different tools for use in various market conditions. Using the STEVE PALMQUIST same tool in all market conditions will likely yield mixed results, because as the market Having the right set of tools and knowing which one to use for a specific task is one of the keys to success in trading Knowing the type of market conditions that are best for each tool requires careful analysis and extensive back testing. environment changes there will be periods when it will work and periods when it shows losses. Knowing the type of market conditions that are best for each tool requires careful analysis and extensive back testing. Trading with a tool that you have not fully analyzed and understand

2 May 2006 is like flying a plane without training. Few people would try to fly a plane without first having someone teach them how it s done. The same March 2002 should be true for trading, but too often that is not the case. This article outlines some of the testing I go through when looking at a potential trading system. Additional trading information and tips may be found at Basic System Requirements Trading systems need to show two key characteristics before being considered for actual use. The system should have a higher percentage of winning trades than losing trades. The average gain of the winning trade should be larger than the average loss of the losing trade. If a trading system shows profitable trades more often than losers, and the average winning trades gain more than the average 2 Table 1. Definition of MAX System Set Up: 1. The Close must be above the 50-day simple moving average. 2. The 15-day simple moving average must be above the 50-day simple moving average. Trigger: The 5-day simple moving average crosses from beneath the 15-day simple average to above it. Entry: Enter after the trigger at the next day s open. Exit: losing trades lose, then the system has a positive expectancy and is something that can be profitable in the long run. One way to determine if your system has a positive expectancy is through back testing. Back testing also allows you to better understand how your system behaves and what filters improve or reduce results. Back testing also answers key questions such as: Figure 1. Daily chart of AmeriCredit with Volume displayed in indicator window. ACF MAX Trigger occurred on 01/04/2006 as shown by arrow. Hold for three days then sell at the next day s open. How does the system behave in different market periods? How is the system affected by stock price or volume? How does varying the system s basic parameters affect results? Basic Multiple Moving Average Crossover System Many traders include in their tool boxes some version of a Multiple Moving Average Crossover (MAX) system. The system tested in this article uses three simple moving averages. The 50-day average is used as a basic filter. The relationship between the 5-day and 15-day average triggers trades. The exit strategy is simply to hold for three days then sell. This results in a simple system that does not require the trader to monitor the market all day. MAX is defined in Table 1. Figure 1 shows a MAX set up for ACF on 01/04/2006. On January 4, ACF is above the 50-day average (red line) and the 5-day average (green line) crosses from below the 15-day average (purple line) to above it. This moving average cross triggers a MAX trade. Over the next three trading sessions, ACF moves up four percent. Four percent in three sessions won t make you rich, but it will put food on the table.

3 May 2006 Initial Back Testing Results The first step in the process of determining if MAX is suitable for trading is to test it in an environment in which it should work. If MAX does not show promising results in a bullish market environment, then there is no point in evaluating it further. The initial testing period for MAX was 8/13/04 through 12/06/ 04, which was a relatively bullish period in the market as shown in Figure 2. Back testing MAX during the market period shown in Figure 2 using AIQ s Expert Design Studio (EDS) software yields interesting, but not compelling, results as shown in Figure 3. The system meets the two criteria outlined above because it shows 54% winning trades and the average winner gains 3.48% while the average losing trade loses 2.76%. Note that the annual ROI numbers shown in back test results assume the trader takes all the indicated trades. This is something Figure 2. Daily chart of NASDAQ with Volume shown in lower window. Time period displayed is August to December 2004, a relatively bullish market environment. that may be impractical due to account size restrictions. However, the annual ROI numbers serve as a good figure of merit for a system; more is generally better. Figure 3. Initial MAX back testing results for the bullish period 8/13/04 to 12/06/04. Analyzing System Filters Running the back test several times in the same time period and using a volume filter for the stocks selected yields the results shown in Table 2 (page 4). Increasing the minimum 20-day average trading volume from 300,000 shares a day to 1,300,000 shares a day had a minimal effect on the percentage of winning trades and the average gain of the winning trades. Increasing the average trading volume requirement reduced the number of trades without providing a financial benefit, so it is not something to include in the MAX system. One of the basic parameters of MAX is the filter requirement that the stock be at least 5% above its 50-3

4 May 2006 day simple moving average. Table 3 shows the results of varying this requirement during the 08/13/04 to 12/06/04 testing period. The March 2002 distance between the stock price and its 50-day average is referred to as daylight since the larger this distance is the more white space (or daylight) there is between the stock chart and its 50-day average. Increasing the amount of daylight in the MAX scan has little effect on the results, other than reducing the number of trades, until the stock is required to be at least 10% above the 50-day moving average. At this point the average annual ROI begins to increase and continues increasing as the daylight requirement becomes larger. Increasing the daylight requirement does not significantly alter the percentage of winning trades. The ROI improvement results from an increasing spread between the average gain of winning trades and the average loss of losing trades. This is an attractive filter that traders should consider incorporating into the basic MAX system. Table 4 illustrates the effect of requiring that the key moving averages used in MAX have positive slopes. Filtering stocks based on a positive slope of the 15-day average dose not improve the system. Requiring that the slope of the 50- day moving average be positive yields a slight improvement in results. 4 Table 2. Effect of Stock Volume on MAX Testing Results Avg. Winner Gain/ Minimum Volume Annual ROI % Winners Avg. Loser Loss # Trades None / , / , / , / , / ,100, / ,300, / Testing In Different Market Conditions: To this point all testing has been done during the 08/13/04 to 12/06/ 04 bullish market environment. The results are interesting and must be confirmed in other bullish periods, as well as tested in other market environments. If a system tests well in one bullish period and not others, then it is possible that the system has simply curve fit the data. If it tests well in several bullish periods it may form the basis of an interesting system. Table 5 shows the results of testing MAX in three bullish market periods, two bearish market periods, and two trading range markets. MAX performs well in all three bullish periods. The percentage of winning trades Table 3. Effect of 50ma Daylight on MAX Testing Results Avg. Winner Gain/ 50ma Daylight Annual ROI % Winners Avg. Loser Loss # Trades None / % / % / % / % / % / % / There are two effective techniques for limiting MAX to relatively bullish market periods: moving average filters and trendlines. averages 60% and the average winning trade gains more than the average losing trade loses. MAX has potential in bull market periods. MAX loses money in both bear market periods shown in Table 5, as one would expect. In bear markets the winning percentage drops well below 50%, and the average losing trade loses more than the average winning trade wins. Either condition would make the system uninteresting in bear markets. Few, if any, trading systems work well in all market conditions. For this reason I have several different systems that have been tested in different market conditions, and I use the ones that are best suited to the current market environment. More information on adapting trading style to the current market conditions is available at The results for the two trading range market test periods are encouraging. The results themselves are only mildly interesting and are not strong enough to result in a practical trading system focusing on trading range markets only. However, since bullish market environments often turn into

5 May 2006 trading ranges prior to a bearish environment, showing mild profits in a trading range environment implies that a filter for determining market conditions does not have to be able to pick exact tops and bottoms. Back testing Max during the five-year period of 01/04/00 to 12/ 06/04 yields 52% winning trades, and an average annual ROI of 48%. During this same period the NASDAQ dropped 44% from 3901 to Adding the daylight and positive 50-day moving average slope to the basic MAX and testing during the same period increases the ROI to 82% with 51% winning trades. The average winner gains 6.84% and the average loser loses 5.19%. This indicates that MAX has some promise since it yielded positive results during difficult market conditions. The data shown in Table 5 implies that these results could be improved by limiting trading to bullish periods in the market and remaining in cash, or using another technique, during bearish market periods. Table 4. Effect of Moving Average Slopes on MAX Testing Results MA Slope Avg. Winner Gain/ Requirement Annual ROI % Winners Avg. Loser Loss # Trades None / ma> / ma > / Limiting Trades to Bullish Market Environments: There are two effective techniques for limiting MAX to relatively bullish market periods: moving average filters and trendlines. Traders can limit MAX trades to periods when the NASDAQ is above a key moving average, or draw trendlines on the NASDAQ and only trade MAX when the market is above an ascending trendline. Back testing MAX during the same 5-year period noted above and limiting trades to periods when the QQQQ was above its 30-day moving average significantly improved results. The average annual ROI increased to 123%, the percentage of winning trades increased to 52.7%, and most importantly the spread between the average gain and the average loss widened. The average winning trade returned 7.18% and the average losing trade lost 4.99%. Table 5. Effect of Market Conditions on MAX Testing Results Market Period Avg. Winner Gain/ Annual ROI % Winners Avg. Loser Loss # Trades Bull Market Periods: 08/13/04-12/06/ / /12/03-09/08/ / /16/03-01/26/ / Bear Market Periods: 03/11/02-10/10/02 <54.37> / /24/01-04/04/01 <42.82> / Trading Range Market: 09/08/03-12/17/ / /05/04-07/02/ / Summary Success in trading, like many other professions, depends on having the right set of tools and knowing which one to use for a specific task. Just as pilots have multiple instruments, traders need different tools for use in various market conditions. If you would like to see my analysis of the current market conditions and the setups I am currently watching, me at max@daisydogger.com. Back testing is a useful tool for understanding how a potential system behaves in different market conditions and with various filters. Understanding how a system behaves in different markets and with different filters is a key part of determining whether or not the system is useful. It is important that traders fully evaluate any system before using it, and clearly understand how it may perform in different market environments. MAX has the potential to be one of the tools in the trader s tool box. During bull market conditions the system produces trades that win more often than they lose, and the average winning trade gains more than the average losing trade loses. The basic MAX system can be improved by adding filters for daylight and a positive slope on the 50-day moving average. MAX trades can be limited to favorable environments by using the 30-day moving average QQQQ as a filter. 5

6 May 2006 Expert Design Studio Techniques Manipulating March 2002 Pre-Built Rules Can Broaden Your EDS Trading Systems By David Vomund T he Expert Design Studio (EDS) package was designed to be easy to use, allowing users to build their own trading systems by cutting and pasting pre-built screening rules. There are cases, however, when you want to do a bit more manipulation of the rules when developing a model. For instance, some people want to include market timing rules in their stock selection systems. EDS can do that. Some people want to screen for stocks that have recently passed a rule rather than simply listing the stocks that pass the rule on the current day only. EDS can do that. In this article, we ll explain how to accomplish these tasks and we will also show you how to build an EDS report. Incorporating Market Timing Nearly all of the EDS models that we ve discussed in past issues of the Opening Bell are pure stock By incorporating market timing code in EDS, you can test various market timing systems and see how these systems affect the overall performance compared to stock screening models. screening models. If a stock passed the screening it was purchased. However, many people want to incorporate market timing in their trading. When the market is poor, 6 no long selections are analyzed. EDS has the ability to incorporate market timing rules. Market timing rules can be created based on any of AIQ s technical indicators applied to a market index. For this article, we ll use the SK-SD indicator and the S&P 500 index. Our rule states that we only want EDS to analyze stocks when the SK line, the faster of the two lines, is above the SD line. Figure 1 shows that a backtest of this rule will have us in the market about half the time. Here is the EDS code for our market timing system: rule1 if [sk]>[sd]. rule2 if tickerrule( spx,rule1). In this simple code, the first rule defines the market timing indicator and the second rule allows the indicator to be run on the S&P 500 (ticker SPX). If you want the market timing model to use a different indicator, then you would modify rule1. For example, if you only want to run an EDS system when the S&P 500 s Velocity indicator is above zero, then the code for the first rule would be: rule1 if [velocity]>0. If you want to run an EDS system when the S&P 500 s Phase indicator is increasing, then you would change the code to: DAVID VOMUND rule1 if [phase]>val([phase],1). Some people want to include market timing rules in their stock selection systems Some people want to screen for stocks that have recently passed a rule rather than simply listing the stocks that pass the rule on the current day only. EDS can do these tasks. To incorporate this market timing code into a stock screening model, simply include rule2 as part of the final screening rule. By incorporating market timing code in EDS, you can test various market timing systems and see how they affect the overall performance of a stock screening model. Screening for Recent Activity When you build an Expert Design Studio model, the screening techniques are usually based on the activity of the most current data. There are times, however, when you may want to allow a stock to pass the screening if the rule has fired within a fixed period of time. That will produce more trades than if we state that the rule must fire on the current day only.

7 May 2006 Let s look at an example. Suppose we want to screen for stocks whose Positive Volume Index (P- Vol) indicator has crossed above its Signal line. We would use the following code: rule1 if val([p-vol],1) < val([p-vol esa],1) and [p-vol] > [p-vol esa]. The first part of this rule states that the P-Vol indicator was below its exponentially smoothed moving average yesterday while the second part of the rule states that the indicator is above its exponentially smoothed moving average today. As we build a trading system we may find that a rule such as this eliminates too many trades. When this is the case, we can change the model to allow stocks to pass the screening rule anytime in the last five days. In our example, we can screen for stocks whose P-Vol indicator moved above its exponentially smoothed moving average anytime in the last five days. Here is the code: rule1 if val([p-vol],1) < val([p-vol esa],1) and [p-vol] > [p-vol esa]. last5days if scanany(rule1,5). Figure 1. Chart of S&P 500 with SK-SD indicator displayed in lower window. Market timing rule allows long buys only while the SK line (lighter shade) is above SD line. In EDS we can insert a last5days folder and run the scan on our database of stocks. A list of the stocks that pass the screening will be added to this folder. We can also add a User Defined Figure 2. Expert Design Studio window with contents of Last5days folder displayed. Symbols of all stocks that passed the screen are listed along with the date when the screen criteria was met. Function that finds the date the rule was passed. To do this, we add the following code to the EDS model: date is ruledate(). After this rule is added, go to EDS and click on the last5days folder. On the title bar, right-click on Symbol and select the Insert choice. Choose User Defined Functions, highlight the Date rule and click OK. With this completed, you can scan your database of stocks and this model will give you a list of stocks that had their P-Vol indicators rise above their moving averages and it will tell you the date of each occurrence (Figure 2). Creating an EDS Report In the previous section, we briefly described how to manually add a date column to our EDS model. In this section, we ll expand the columns and build a report that displays more information on the stocks that pass the screening. Even better, we can accomplish this by copying and pasting rather than manually typing. 7

8 May 2006 Your AIQ software comes with a file that contains several pre-built User Defined Function (UDF) rules. To open this file, go to EDS and March 2002 click File and then Open. Doubleclick the EDS Strategies folder. Next, double-click the Pre-Built Functions Demos folder. Highlight the How To Use Tickerrule file and click OK. This file contains several choices of pre-built UDFs. A rule is a UDF if the second word in the rule is is. In this file we chose to copy the Close, Group, and Sector UDFs to the clipboard. These UDFs can be pasted into any EDS file that you use. For this example, we pasted the UDFs into our P-Vol model. With these UDFs placed into the EDS model, we highlighted the folder that contained the final EDS screening rule (the last5days rule). Next, highlight Symbol on the title bar and right-click. Select Insert, choose User Defined Functions, and then individually select the UDF rule names. Your EDS screening will now appear in a report format, showing the name of each stock that passes Newsletter David Vomund, Publisher G.R. Barbor, Editor P.O. Box 7530 Incline Village, NV Figure 3. Expert Design Studio window with contents of Last5days folder displayed. Note that four information fields have been added by pasting UDF's into the original screening rule file. the screening along with its closing price, industry group, and industry sector (Figure 3. With the report created, you can click on any column title to sort the data by that column. In addition, you can move the file to a spreadsheet program by clicking Report, Export, and then save the file using a csv format. As we build a trading system we may find that a rule eliminates too many trades. When this is the case, we can change the model to allow stocks to pass the screening rule anytime in the last five days. David Vomund publishes VIS Alert, a weekly investment newsletter. For free trial issues, call or go to does not intend to make trading recommendations, nor do we publish, keep or claim any track records. It is designed as a serious tool to aid investors in their trading decisions through the use of AIQ software and an increased familiarity with technical indicators and trading strategies. AIQ reserves the right to use or edit submissions. While the information in this newsletter is believed to be reliable, accuracy cannot be guaranteed. Past performance does not guarantee future results , AIQ Systems Check Out Back Issues of the Opening Bell Newsletter For a 12-year archive of back issues of the Opening Bell newsletter, visit: You can also search for particular topics on this site 8

9 May 2006 Market Review Trading Range Continues Breakout Due T he market averages were mostly flat in April. For the month, the S&P 500 gained 1.2% while the Nasdaq fell 0.7%. Mid-month there was volatility. The S&P 500 fell below its 1290 support level. The first day of this sell-off was an outside reversal day. An outside reversal takes place when the day s high is greater than the previous day s high and the day s close is below the previous day s low. Nevertheless, the S&P 500 s overall pattern of higher lows persisted and the pullback was short-lived. By month's end, the market was in a very tight trading range. For the S&P 500, the upper limit was 1313 and the lower limit was To demonstrate the lack of a trend, the ADX/ADXR indicator, which measures the strength of a trend is near new multi-month lows. The range is so tight that a breakout should happen soon. If the breakout is to the upside, then 1313 will turn into support. If the breakout is to the downside, then 1290 becomes resistance. As always, some groups performed especially well in April even with a flat market. Precious Metals rose 12%, Energy rose 7%, and Banking rose 5%. The losing groups were Biotechnology and Health Care, each losing about 5%. While the U.S. market stalled in April, many international markets remain red hot, and Exchange Traded Funds (ETFs) are great trading vehicles to take advantage of this strength. Latin American is the hottest and can be traded with ishares Latin America (ILF). STOCK DATA MAINTENANCE The following table shows stock splits and other changes: Stock Ticker Split/Div Approx. Date Zevex Int l ZVXL 3:2 05/05/06 Progressive Corp. PGR 4:1 05/09/06 Norsk Hydro ADS NHY 5:1 05/15/06 Anadarko Petroleum APC 2:1 05/30/06 Talisman Energy Inc. TLM 3:1 05/31/06 Barnes Group Inc. B 2:1 05/31/06 Trading Suspended: Abegenix Corp. (ABGX), Burlington Coat Factory (BCF), Burlington Resources (BR), Chiron Corp. (CHIR), Hughes Supply (HUG), Jefferson Pilot (JP), Maytag Corp. (MYG) Name Changes: American Pharmaceutical Partners (APPX) to Abraxis BioScience Inc. (ABBI) CNF Inc. (CNF) to Con-Way Inc (CNW) Fortune Diversified Industries (FFI) to Fortune Industries Inc. (FFI) Frontier Airlines (FRNT) to Airlines Holdings Inc. (FRNT) United Mobile Homes (UMH) to UMH Properties Inc. (UMH) Popular AIQ Seminars Free Web Seminars AIQ offers free online seminars for both beginning and experienced AIQ users. For May s schedule of these popular seminars, please visit One-Day Seminar San Diego, CA May 20 Topics include the use of TradingExpert Pro s more important features plus trading techniques and market timing. For more information go to S&P 500 Changes Changes to the S&P 500 Index and Industry Groups: Legg Mason (LM) replaces Guidant Co. (GDT). LM is added to the Banks (BANKGRP) group. SanDisk Corp. (SNDK) replaces Chiron Corp. (CHIR). SNDK is added to the Computer Storage & Peripherals (COMPUTEP) group. Dean Foods (DF) replaces Maytag Corp. (MYG). DF is added to the Packaged Foods (FOODCOMP) group. Kimco Realty Corp. (KIM) replaces Jefferson Pilot (JP). KIM is added to the Real Estate Investment Trust (REITRUST) group. Boston Properties (BXP) replaces Cinergy corp (CIN). BXP is added to the Real Estate Investment Trust (REITRUST) group. Google Inc. (GOOG) replaces Burlington Resources (BR). GOOG is added to the Internet Software & Services (NETSFTWR) group. 9

10 May 2006 AIQ Seminars 2006 One March Day 2002 Seminars Across the Country with David Vomund and Steve Palmquist $288 Here's what previous attendees had to say "Palmquist was great! A nice combination of good systems and good methodology clearly presented in an engaging manner" "David Vomund. Very knowledgeable, very good speaker, extremely practical advice-a great asset to AIQ!!" VENUES San Diego, CA May 20, 2006 AIQ's One Day Seminar Agenda Offers Top Class Education at a Great Price Short Term Trading Techniques for Trading Both Long and Short Positions by Steve Palmquist In this session Steve will be sharing one of his favorite short term trading techniques for trading both Long and Short positions. This is not a theoretical approach, but one Steve uses every day with his own money. He will discuss the specific system, show how it was developed and tested, and provide tips and techniques for using it based on his nearly 20 years of market experience. Market Truths by David Vomund David Vomund will reveal important market truths that he has learned from actively studying and trading the market for nearly 20 years. These are the truths that experienced investors learn over time. New investors learn these truths the hard way by losing money. Topics include market behavior, position sizing, and trading psychology. Market Adaptive Trading by Steve Palmquist Steve has developed a Market Adaptive Trading Technique that he uses to determine when to trade specific systems and when to stand aside. Successfully analyzing current market conditions is one of the keys to trading success. Learn when to focus on Longs, when to trade Shorts, and when to stay in cash. Just trading a single system all the time without a careful analysis of the market conditions can give you a lot of practice at taking draw downs. Learn from Steve s experience and take your trading to the next level. Getting Started Right With AIQ by David Vomund David Vomund will cover the features within TradingExpert that every AIQ user should know. Basic features such as charting, report generation, and exploring will be discussed. Advanced features such as industry group creation and the use of Expert Design Studio will also be covered. Sign up now x115 10

11 May 2006 AIQ Seminar Speakers David Vomund Chief Analyst AIQ Systems David Vomund is editor of AIQ's Opening Bell newsletter, and publisher of VISalert.com. Through his investment advisory firm, Vomund employs The Style Index Strategy on client portfolios ( He was published in Active Trader, Technical Analysis of Stocks & Commodities, and was a contributing author of Computerized Trading - Maximizing Day Trading and Overnight Profits. Steve Palmquist is a full-time trader and publisher of daisydogger.com. Steve uses AIQ's Expert Design Studio to find interesting trading ideas and strategies. He shares his trading techniques and systems at seminars across the country, including the Traders Expo and AIQ seminars. Steve has been published in Stocks & Commodities, Active Trader, and Working Money as well as AIQ's Opening Bell. Steve Palmquist Founder Daisydogger.com Limited Seating Sign up Now 11